Determing if ChIP Signal Is Real - (Jan/26/2010 )
So I've got a general question concerning how to figure out if one's signal from ChIP is real or not. I know this is a long post, but I suspect a discussion of these issues would be useful to a number of folks...
There appears to be quite a bit of variability in the literature in terms of how ChIP data is presented; from %input to specific/mock to just reporting the comparison of a control group to an experimental group. Thus I (and I imagine others) struggle to understand what a "real" ChIP signal is.
From an initial reading of the literature it seemed that specific/mock (e.g, IgG) would be a good way to determine if what one is seeing is real. However, the primary drawback I see to this is how to deal with highly variable mock data (At least that's what I get, sometimes it doesn't even show up after 40 cycles in some replicates, which makes it difficult to quantify (score it 40? throw it out?...)). However, in the only two published articles I've found that actually show their raw qPCR data the Mock data looks fairly clean and quite high! (See Hao, Liu, Gonye and Schwaber (2008) J Neurosci Meth 172, 38-42; and Nelson JD, Flanagin S, Kawata Y, Denisenko O, Bomsztyk K (2008) Am J Renal Physio 294, F525-33 in supplemental materials).
Is the Mock data in these articles (Cts around 25-30) what Mock data should look like? Or should it be difficult to get any signal?
The other suggestion that has been mentioned on this board is to use primers towards a genetic region that is not transcribed or does not bind the protein of interest. This approach makes a lot of sense to me, however it remains a bit more difficult to do than it seems on the surface. For example, I'm working with brain tissue (the hippocampus) so I figure using primers for rhodopsin as well as those aimed at clusters of olfactory receptors should be a safe bet in terms of genetic regions that should be shut down in an area that does not directly detect light or odors. However, upon doing ChIP for RNAPII (using 2ug of DNA, incubating overnight using Magna ChIP kit from Millipore) I find that I get a %input of these regions that is on par with my gene of interest as well as other genes that I know are transcribed in this brain region! However, the fold change over my mock IgG sample is low for rhodopsin and olfactory receptors (typically < 10 whereas in genes known to be transcribed it hovers around 20-60). BUT of course there is significant variability in the Mock IgG samples, so can this data be trusted? If %input values are similar but specific/mock is consistently different, does this mean these are good negative control regions? If not, anyone have a good suggestion for a negative control region? I'm thinking perhaps one solution is to find a way to increase my qPCR signal and thereby get cleaner Mock data like in the publications mentioned above? Perhaps by adapting the DNA purification of the FastChIP protocol?
Well I guess I'll leave it at that for now. Thanks for reading this far! Any insight is appreciated.
As a follow-up to this, I noticed in one of the articles (Hao et al) they first amplified their ChIP product for 15 cycles in a thermal cycler, then removed 2.5 uL of the PCR product and did their qPCR on that. Has anyone tried this? Is this a way for reducing variability in the qPCR data due to low amounts of starting DNA? I'm not sure I see how this would be beneficial over simply running the qPCR for more cycles to amplify small amounts of DNA.
Well I am relatively new to ChIP per se but not molecular biology. For me personally, I think the IgG control is not much of a control to speak of really. I use the Magnify ChIP kit from Invitrogen and there is really no background signal from IgG at all, or you get a primer-dimer type signal (>35 Ct) which, in my opinion, is probably not useable. My first reaction was to score them all at Ct 40 for the sake of calculations and I think this makes sense. If I got a Ct of 20 - 25 for my IgG control I would be worried!
If you perform PCR on a negative control region and see a significant difference in % input then I think you are good to go. Like you I too saw a weak signal for Pol II at a gene that should not be expressed in my tissue of interest. However since my positive control region gave a stronger signal, I was fairly happy with it. I have also redesigned the primers a bit to move them to a more 3' intronic region (they were previously in an exon!). Regardless of the signal, intuitively this is a much better control and is a lot more rigorous......which is why I am sure many will turn away from it.